Tidy

What Is Tidy?

Tidy is a Claude Code skill designed to streamline the organization and categorization of financial transactions, particularly those that remain uncategorized in your financial data. Triggered by natural language commands such as "tidy up", "clean up transactions", "categorize uncategorized", and "organize my transactions", Tidy automates the process of sorting through large volumes of transaction data. By leveraging clustering techniques and intelligent research methods, Tidy helps users batch-categorize similar transactions efficiently, reducing manual overhead and improving financial data accuracy. The skill is a productivity enhancer, especially valuable for individuals or businesses managing extensive transaction histories.

Why Use Tidy?

Uncategorized transactions are a common pain point in personal and business finance management. They obscure spending patterns, compromise reporting accuracy, and often require tedious manual review. Tidy addresses these challenges by:

• Automating tedious work: Instead of manually reviewing each transaction, Tidy groups similar entries, making bulk categorization possible.
• Improving categorization accuracy: By researching unknown merchants (via web and email search), Tidy increases the likelihood of accurate assignments.
• Supporting better financial insights: Clean, categorized data enables more reliable budgeting, forecasting, and compliance reporting.
• Reducing errors: By presenting suggestions in clusters, users can review and approve changes with confidence, minimizing the risk of misclassification.

Organizations and individuals seeking to maintain up-to-date, well-organized transaction records will find Tidy especially beneficial as their transaction volume grows.

How to Get Started

To use Tidy, you need a setup that supports Claude Code skills and access to the Tidy skill source. The core workflow consists of several well-defined steps, which can be invoked using natural language commands.

Example: Triggering Tidy via Command

You can activate Tidy with commands such as:

tidy up
clean up transactions
categorize uncategorized
organize my transactions

Technical Workflow:

1. Fetch Uncategorized Transactions

   Tidy queries your transaction database or API to fetch uncategorized transactions, typically from a recent period (e.g., last 90 days). You can specify a period as needed.

   json

   {
     "detail": true,
     "is_uncategorized": true,
     "period": "last_90d",
     "limit": 200,
     "sort": "-amount"
   }

   To customize the period, simply include it in your command (e.g., "tidy up transactions from last month").

2. Research Unknown Transactions

   For any unfamiliar transaction descriptions, Tidy attempts to identify the merchant using:

   • Web search: Looks up merchant names, phone numbers, or transaction descriptions.
   • Email search: Scans the user's email for order confirmations or receipts matching the transaction description or amount.

3. Cluster by Pattern

   Tidy groups transactions by normalized descriptions or party names. For example, transactions with similar descriptors (possibly with minor variations) are clustered together for bulk processing.

4. Suggest Categorization

   For each cluster, Tidy proposes a category (from your existing set) and a cleaned-up merchant or party name.

5. Present Results

   Finally, Tidy provides a summary table showing the proposed categorizations, counts, and total amounts for each cluster, making review and approval straightforward.

Key Features

• Batch Categorization: Categorize hundreds of transactions at once by clustering similar entries.
• Intelligent Research: Utilizes web and email search to accurately identify merchants behind cryptic transaction descriptions.
• Customizable Time Periods: Query transactions from any specified period, such as "this month" or a custom date range.
• Automated Grouping: Clusters transactions by normalized party names or descriptions, reducing duplication of effort.
• Category Suggestions: Suggests categories based on your existing transaction data, ensuring consistency.
• Review Table: Presents a clear, actionable table for user review before finalizing any categorization changes.

Practical Example:

Suppose your uncategorized transactions include the following:

Tidy would cluster both Amazon transactions, propose the "Shopping" category, and clean "AMZN Mktp US*" to "Amazon". The Starbucks entry would be placed in the "Dining" category with the party "Starbucks".

Best Practices

• Review Clusters Before Applying: Always inspect Tidy’s suggested clusters and categories to ensure accuracy, especially for ambiguous transactions.
• Update Categories Regularly: Keep your list of categories current to maximize Tidy’s usefulness when suggesting assignments.
• Utilize Time Period Filters: Specify periods in your command (e.g., "last month") to focus on recent or relevant data.
• Augment with Manual Corrections: For edge cases or personal spending patterns, manually adjust categories as needed before approving changes.
• Leverage Email Integration: If available, authorize Tidy to search your email for receipts to further improve merchant identification.

Important Notes

• Privacy Considerations: Tidy may access transaction descriptions and, if authorized, perform searches using your email content. Ensure you comply with your organization’s privacy policies and only grant access where appropriate.
• Limitations: Tidy’s merchant identification is only as accurate as the available data and search capabilities. Some transactions may remain ambiguous, requiring manual review.
• Batch Limits: The default query limit is 200 transactions per batch. For larger datasets, consider running Tidy in multiple passes.
• Category Consistency: Tidy only suggests categories from your existing set. New categories must be created separately.
• Skill Updates: Refer to the official Tidy repository for the latest features, improvements, and bug fixes.

By integrating Tidy into your financial workflow, you can significantly reduce the time and effort spent on transaction organization, leading to cleaner, more actionable financial data.